Date of Award
Doctor of Philosophy
The objective of this research is to discover the search state and path patterns through which users retrieve information in hypertext systems. The Markov model is used to describe users' search behaviour. As determined by the log-linear model test, the second-order Markov model is the best model. Search patterns of different user groups were studied by comparing the corresponding transition probability matrices. The comparisons were made based on the following factors: gender, search experience, search task, and the user's academic background. The statistical tests revealed that there were significant differences among all the groups being compared.;A three-way analysis of variance test was conducted to study the effects of gender, search task, and search experience on search option (analytical vs. browsing), as measured by the proportion of nodes reached through analytical searching. The search task factor influenced search option in that a general task caused more browsing and a specific task more analytical searching. Search experience alone did not affect the search option. There was a possible gender difference among high experience users, with males favouring analytical searching more than females.;Two frequency distribution models were developed and tested to describe path patterns. Path length followed a shifted negative binomial distribution. The frequency of node visiting followed a Zipf distribution.;The resulting probabilistic models can help us better understand users' search behaviour and the search process involved. They provide valuable information in evaluating a system's existing operation and in refining future design. They also provide a background for examination of systems via simulation studies.
Qiu, Liwen, "Probabilistic Models Of Search State And Path Patterns In Hypertext Information Retrieval Systems" (1991). Digitized Theses. 2013.